计算机应用与软件2017，Vol.34Issue(11)：295-301,320,8.DOI:10.3969/j.issn.1000-386x.2017.11.054
基于随机最小冗余条件互信息和支持向量机的混合入侵检测特征选择
FEATURE SELECTION OF INTRUSION DETECTION BASED ON RANDOM LEAST REDUNDANT CONDITIONAL MUTUAL INFORMATION AND SVM
摘要
Abstract
There are a lot of irrelevant and redundant attributes in intrusion detection logs,which seriously affect the detection of real-time.Most feature selection algorithms cannot take into account the correlation and the amount of information,and are easy to fall into the local optimal solution.To solve this problem,a hybrid intrusion detection feature selection method based on random minimum redundancy conditional mutual information and support vector machines (SVM) is proposed.Firstly,mutual information and correlation were used to eliminate redundant features without classification information and a high degree of correlation between features.Then,the random minimum redundancy condition mutual information criterion and SVM were used to select the optimal feature subset with the maximum classification information quantity,avoiding the local optimal solution to a certain extent.Our results show that it can get an optimal minimum feature subset effectively on condition of ensuring the intrusion detection classification accuracy.关键词
信息熵/互信息/支持向量机/特征选择/入侵检测Key words
Information entropy/Mutual information/SVM/Feature selection/Intrusion detection分类
信息技术与安全科学引用本文复制引用
丁宣宣,郭渊博..基于随机最小冗余条件互信息和支持向量机的混合入侵检测特征选择[J].计算机应用与软件,2017,34(11):295-301,320,8.基金项目
国家自然科学基金项目(61501515). （61501515）
